Numerical simulation and ANN prediction of phase change material embedded within 3D printing lattice structures
Suping Shen
Abstract
This study investigates the phase change material (PCM) embedded within different lattice structures, including SC, BCC and FCC, at various porosities. The examination focuses on understanding the thermal behaviour and heat transfer characteristics during the melting and solidification processes. Key parameters analysed include the maximum temperature of the heater, PCM melting and solidification, and Nusselt number. The results indicate that the maximum temperature of the heater decreases with increasing porosity. The Nusselt numbers for the different lattice structures exhibit similarities, with the SC lattice showing a slightly higher Nusselt number. Melting and solidification times increase with increasing porosity. An artificial neural network trained by the Bayesian Regularization algorithm is used to predict three key parameters. The porosity and time are set as input parameters of the network. The optimal structure of the ANN show high accuracy in estimating the performance of PCM with three lattice structures. The minimum mean square error and the maximum correlation coefficient are 0.00003601 and 0.9998, respectively. The trained ANN is employed to predict the behaviour of PCM with 82 % SC, BCC and FCC. A perfect agreement is observed between ANN predictions and the simulation for PCM with lattice structures featuring 82 % porosity.